Military Target Detection by Integrated Spectral Design and Radiative Transfer Modeling

A paradigm shift in modern-day warfare, spectral designing and signature management has gained prominence in the defence industry. From a target detection perspective, various national military establishments are making efforts to develop sensors, signature management, and survivability of their own military platforms. Most targets of strategic importance are found camouflaged in tactical environments. Hyperspectral imaging (HSI) is increasingly used in defence applications at various information levels. Acquisition of spectral signatures of adversaries and discriminating between actual and decoy targets is vital for deployment in meeting time-sensitive intelligence needs. We have developed a novel spectral target detection approach, integrating radiative transfer modelling (RTM), a non-linear spectral mixture model, and a reference spectral library database, to discriminate real and decoy targets of military interest. We acquired spectral signatures of various strategically important artificial as well as natural materials and developed a forward spectral mixture modelling based approach to engineer spectral signatures, which we name the primary decoy signature database (PDSD), conforming to real landscapes. The spatial variability and environmental interference of all the targets are modelled with respect to geometric and atmospheric conditions using RTM. The proposed approach is experimentally implemented for target detection in a HySIS hyperspectral satellite imagery acquired over defence terrain in the Indo-China Himalayan region, conforming to various scenarios of target-material optical perspective. The results suggest operational promise of spectral signature design and engineering for differentiating real and decoy targets using HSI.